Trabajo Fin de Grado - Ingeniería de Software

Facial expressions detection using Keras model and Haar Cascade classifier. Trained with FERPlus and AffectNet with the following results:

Number of classes	Accuracy
3 classes	86%
5 classes	74%

Requirements

You can install the required packages directly from the requirements.txt file using the following command:

pip install -r requirements.txt

You also need the following:

• Python 3.10 (will not work with older versions)
• Webcam for Camera mode
• Linux, macOS or Windows operative system

How to run

Use this command for help:

python3 run.py --help

It will output the next options:

• -h, --help: show this help message and exit
• --class-mode {basic,extra}: Number of classes to detect. Basic mode classifies 3 diferent emotions with 87% of accuracy while Extra mode classifies 5 different classes with 74% of accuracy.
• --input-mode {webcam,screen}: Input source to detect faces.
• --background, --no-background: Execute in background without window.